Within my AD structure, group management is delegated within certain OU’s, I now need to replicate that functionality in the FIM portal.
The is no real way of identifying which groups should be managed by whom, except the OU within which the group currently resides.
So, to start off with I need to get the parent OU of the group into the portal:
Import the OU into the MV:
Setup an export flow for adOU into the portal.
Then, by using the Lithnet PowerShell Module, we can create all the sets and MPR’s required, below is a sample for creating one delegated “collection”. In production, my XML file is much bigger – delegating group management to around ten different groups.
Note, that you first need to create references to all users who might be given the rights to manage groups. This includes the FimServiceAdmin and FimServiceAccount – referenced by their ObjectID, the others are referenced by their AccountName. All members referenced in this section, are added to the __Set:GroupValidationBypassSet. This set is defined in the non-administrators set – not in this set – this bypasses the group validation workflow:
Create a set of groups to be managed – the filter being the OU that the groups belong to & MembershipLocked=False
Create a set of administrators for this delegation – adding the explicit members
Then create the two MPR’s to allow the members of the administrative set to manage those groups – the first MPR allows modification (Read, Add and Remove) of the ExplicitMember attribute, while the second allows creation and deletion.
Use Import-RMConfig -File <PathToXML> -Preview -Verbose to validate your xml and see what it would do. Drop the “-Preview” to make the change
This seemed to me to be a really nice solution, just defining all exceptions and suffixes within one file, read it in on code execution, then check for existence or whatever in the code.
So, given the following xml file:
Add the System.Xml Import and declare the variables, so they are global:
Add the code to read the xml file into the Initialize Sub:
Then, when you wish to look for those values within those variables – just like in the last post:
While I was happy that this method worked, I was not entirely happy with the fact that I still had some hard coded values in the code. However, the way that the function operated meant that if I took my collection of mail suffixes (20+) and added them all to the text file, then the array would be built for each and every user that passed through the dll, not too efficient!
So, I was looking for something a little more elegant. I was happy for the array to simply be defined when the dll was loaded.
Here is my solution:
At the beginning of my AD MA, I declare my dates and logging levels etc, then generate those arrays using the function. These arrays are now static and are good for processing all users without being regenerated.
When I wish to look into the array to validate a valid email suffix for example, I go from this (as in the last post):
To this:
Much cleaner – plus all suffixes can now just reside in a text file.
Note that updates to the text file will only be realised if the dll is reloaded and the array is regenerated. I believe that this is after 5 minutes of inactivity and seems to hold true from testing.
A long term goal of mine, has been to get “account requestors” to take ownership of their Service Accounts.
Attempts have been made by my predecessors to record an owner of a service account, but it has simply been done as a string attribute of the AD object. Thus, when the person leaves and the account is deleted, the service account becomes orphaned, with an reference to a long forgotten ID.
So thinking of a way to carry this out….. I am already using the email address of the owner of an administrative account to make decisions about whether the administrative account should be enabled or disabled – based on the end date of the owner – discovered by looking up the email address in the MV.
I figured that I could do something similar for those Service Accounts. I’ll be creating service accounts via the portal, the owner of the account will be assigned to the manager attribute. So, how can I get the email address of the manager into the MV as a thing that I can lookup??? I can’t do an advanced flow rule on the FIMMA, and even if I could, Manager is a reference attribute, so I can’t do it anyway… I found an article about dereferencing another attribute, that get me going down this path….. The solution is simple. Create a new attribute and binding in the portal – “ManagerEmailAddress”, then setup a workflow as follows:
When the account falls into scope, the managers email address is set into that new attribute – in the sync engine create a direct flow to put that into the MV (I’m using “serialNumber” – for one reason or another, that I wont go into :)).
I have on the import from AD, some code to set an MV boolean flag – “functionalID” – if the DN of the person object contains the strings found in the Service Account OU’s, thenfunctionalID = True. This attribute is pushed into the portal and is used in set definitions.
So, I’m getting there. Now I need something to set another flag in the MV that will go to the portal. this one defines if the owner of the Service Account is approaching their end date (30 days prior):It is defined on the Import from AD and populates the MV attribute “functionalID-owner-expiring”
Of course after initial code definition, I found another of those inevitable exceptions, so added the generateArrayFromFile function, with a reference (in txt file) to the email address that should be ignored.
Create attribute and binding in the portal for FunctionalID-owner-expiring
Setup an Export in the FIMMA for the new attribute
Create a set: FunctionalID = True and FunctionalID-owner-expiring = True.
Create notification workflow and mail template: notification to [//Target/Manager], then the set transition MPR.
I think I have it, just need to do a little testing to see that it works as expected.
I’m still a long way from the stated goal, as I still need to find “owners” for all of those accounts that have been created in the past.
In the last few days, I have had a few more exceptions to cope with in my FIM Config.
Another new mail suffix
A user who is employed by one tenant, who has that tenants email address suffix; but who is on secondment to another tenant, who have a different mail suffix. The users attributes have been changed in the HR system, so that they gain access to the stuff in the other tenant, which is controlled by automatic groups, based on attribute data!
So, I’d been thinking for a while about having a method to add exceptions without having to add them to the code directly and thus forcing a rebuild followed by full syncs. I found a nice function to read a text file to an array, this is added to the top of the dll after the lines:
Public Class MAExtensionObject_YourMA
Implements IMASynchronization
So, the referenced file simply has the email address of the user that I don’t want to be alerted about. If the email address does not match the expected value, look in the array generated from the text file; if it in not in there either raise an error to get this fixed or investigated.
I have a hardcoded list of those that are already in use in the dll, if the suffix is not found in that array, it does a lookup of the array generated from the “suffixes” text file, if it is not in there it raises an error:
When trying something new out with FIM Development, I often see how to do it in a console app beforehand. Then once I have the process/ method worked out, I translate it into FIM code. Usually this is a very clean process and is quicker than editing the FIM code directly, then doing sync’s on individual accounts.
After sorting through the data that I obtained from the collection script, removing duplicates, etc. The result set was slightly messier than I expected, quite a few manually added entries and incorrect ones.
Anyway, to get the bulk of the changes implemented, I’m just targeting the large expected set – those with VPN settings labelled SITE1 PPTP and SITE1 VPN (anonymised…). Those with SITE2 PPTP have the settings, that were previously enforced by another script, have them enforced here too. This script will take the place of the existing one.
New machines, who do not have the settings already have them defined. Anyway here it is:
So this is not directly relevant to FIM per se, but it falls under the kind of IdM/ Authentication umbrella, so I thought it belonged here….
In December 2014, I bought a Yubikey Neo. I wanted to see how it could be used to harden access to some sensitive “stuff”.
These are really cool devices; they are relatively inexpensive (~£36), yet provide a bunch of functionality all on one device, some of which I have not used.
The components that I did use were:
Yubico OTP – the One Time Passcode functionality that is present OOB – used to sign into the Yubico Forums
U2F – I use this for 2FA for my Google accounts and this blog – it is very simple to set up this 2FA method across multiple services. Look here for more information: https://www.yubico.com/applications/fido/
SmartCard (PIV) – this was the part that I was really interested in for securing stuff within the enterprise. I had recently installed a Windows PKI Infrastructure, so used that to generate trusted SmartCard Logon Certificates to install onto the devices. Look here for configuration docs: https://developers.yubico.com/yubico-piv-tool/
As with most of these things the documentation was initially difficult to read, there were various command line tools to manage different aspects of the Yubikeys, some of them had bugs at the time.
Anyway, long story short, back then I got it configured just how I wanted and used it daily ever since. However, just before Christmas, the SmartCard certificate that I had generated the previous year expired. Thus, the SmartCard functionality of the Yubikey became invalid.
I generated myself a new certificate from my CA, then came to try to remove the old certificate from one of my Yubikeys. I could not because I needed to authenticate against the device to carry out this action. The authentication string (aka password) is called the Management Key, that is (should be) changed from the default value when configuring the device. I went on the scrounge trying to find the key for this particular device, I found my old notes (command line dumps) from the previous year, there were a few management Keys within but not one for this particular device.
OK, so now I need a new management key….. The docs use dd to generate the key:
At the time, I didn’t have easy access to a Unix system to do this, but more importantly I wanted to find a way to achieve the same result in Windows, using PowerShell. This would allow me to script the whole process. Here is the script to create the management key (For info about what “{0:X2}” means, look here: http://www.powershellmagazine.com/2012/10/12/pstip-converting-numbers-to-hex/)
I have now re-written a script that I put together last year to add initialise a new or reset Yubikey (with PIV support) and add a user SmartCard certificate from a Windows CA:
While looking to get this working: https://github.com/shield-9/u2f-login/, I was seeing errors in the http logs referring to OPENSSL_VERSION_TEXT. These were being generated by the following lines in the u2f.php file:
After running a few more tests, it seemed that this PHP Constant should have been defined, but was not. Goggling didn’t provide any quick answers, but I soon worked it out for myself.
In the php.ini file (under arch in /etc/php/), uncomment the line: “extension=openssl.so” and restart httpd. Once the extension is enabled, the Constant is then available.
In the end I did not use this plugin, I was unable to register my Yubikeys. The plugin seems to have been superseded by this one: http://github.com/georgestephanis/two-factor/. In order to have the ability to use U2F login using this plugin, the openssl.so extension has to be enabled; otherwise, the U2F login option is not available.
Recently, one of my co-workers was assigned to work with me to learn some FIM and to help, where possible, with some configuration.
One of the things on my list of “things to do” was to find a way of hiding certain portal buttons from a normal users view – after all they are not (currently) going to be allowed to use the portal for any self service task yet – that might come later….
So, I sent my co-worker off to find and implement the solution in the test environment, thinking “It can’t be that hard…”
This was done around 1 month ago and already I’m forgetting the actions that were carried out, so I’m going to document the steps in a more verbose fashion here for future reference.
Create new search scopes, using the original Search Scopes’ values as the basis. The resource type shold be “Resource”.
I had already setup a new Usage Keyword “PrivilegedUsers”, that I was using to control the things that the helpdesk users would see to do their admin tasks. So, in the original search scopes this keyword needs to be added and the BasicUI and GlobalSearchResult keywords should be removed. Then in the (new) copies of the search scopes, BasicUI and GlobalSearchResultshould be present.
Original and New Search Scope Usage Keywords:
Do an IIS reset….
Now on the home page, choose one of those new search scopes and do a “blank” search. Copy the URL that this search went to – note that the UI of this search presents only the details button.
Do the same thing for each of the new search scopes, copying the URL’s for later….
Now we need to make those URL’s available via the portal UI. We need to do this for both the Navigation Bar and Home Page, so create new resources, using the information within the originals as the basis. The only change that I made was to remove the reference to (DGs) and (SGs) for the top level items and to add a description “User Read Only view” for those new items with the same name (so that I could tell the difference):
Use the URL’s that were obtained from the search scopes to define the behaviour of the Resource:
Do an IIS reset….
If you copied the Parent Order and Order values of the original Resources, the result is a little messy from the Administrators POV, as all of those items are crammed together. Unfortunately, I know of no way to hide an item from the admin, so I just rearranged the order of the layout. It displays as follows after re-arranging both the Nav Bar and Home Page Resource orders:
A normal user looking at the portal sees this, the links go to the URL’s in those Search Scopes defined earlier:
And when looking at for example “Distribution Groups”, sees that Details button only:
The PrivilegedUsers view, looks like this:
And if a member of PrivilegedUsers uses the Distribution Groups (DGs) link, they see this – note all of the buttons are available:
If I remember missing something or something not being quite right, I’ll be back to correct…
